Log Buffer #108: A Carnival of the Vanities for DBAs
Aug 1, 2008 / By David Edwards
Welcome to the 108th edition of Log Buffer, the weekly review of database blogs.
With almost no ado at all, let’s begin with the bad news–from StatisticsIO and Jason Massie: The Death of the DBA. And who is the perpetrator of this crime? The Cloud! It sounds like something from a John Carpenter movie, doesn’t it?
Let’s see what Jason is thinking. “I’d like to retire a SQL Server DBA with 40 years experience but I don’t think that will happen. The cloud is coming and it is bad news administrators, database or otherwise. . . . Let’s make some assumptions. The features get there. The availability gets there. The platform basically matures . . . Now put yourself in the IT decision maker’s shoes. No upfront capital expenses, no managing backups, and no patch management. . . . If they can remove their focus from managing and deploying IT, they sell and service more widgets.”
Scary stuff, right? Well, the commenters don’t entirely agree. I think it will be at least a factor, but I wonder how many managers will look at “The Cloud” and feel uncomfortable about privacy, data retention, and the like. (For myself, I couldn’t even endorse the idea of putting this blog’s comments into “The Cloud”.) What do you think?
Elsewhere on StatisticsIO, Jason has a note about MSDN’s SQL Heroes contest, whose aim is to, “. . . create a community project in CodePlex based on SQL Server 2008.” Jason also links to a list of CodePlex’s active SQL Server projects.
Turning to matters technical, Jeff’s SQL Server Blog offers a lesson on converting input explicitly at your client: don’t rely on the database to “figure it out”. Jeff takes the example of formatting dates, and show both the right and the wrong way, writing, “I’ve said it over and over and I’ll say it again: The concept of formatting dates should never be something that your database code should ever worry about.”
On the Less Than Dot blog, SQLDenis observes that converting columns to date from datetime does not result in a scan in SQL Server 2008. What you get instead is a seek, as he demonstrates.
Indexing Foreign Keys – should SQL Server do that automatically? So asks Greg Low on the The Bit Bucket. “By adding indexes on the foreign keys on three tables,” he writes, “we saw a reduction of 87% in total I/O load. . . . it really struck me that having SQL Server do this by default would avoid a lot of apparent performance problems. . . . Should SQL Server simply do this by default when you declare a foreign key reference?”
Kent Tegels of Enjoy Another Sandwich — riddle me this, riddle me that! “When is a bug not a bug?” I give up, Kent. When is a bug not a bug? “I guess when the developers decide it is not.” Ahhh. Anyway, Kent believes he found one in SQL Server’s spatial/geo implementation, and filed a report on it. “The response that came back was authoritative and well-reasoned. Yet it also smacks of ‘its not an error because we say it isn’t.’ Hopefully this response (and kind of response) will be reconsidered in the future.”
SQLBI’s Marco Russo’s bonnet got a visit from the same bee. He rants a little about the attitude, “It’s not a bug, it’s by (bad?) design”: “If you pretend that something is a bug, you are stating that a developer wrongly coded a program because it does not meet the requirements and the specifications. If the design is wrong, and the program is written ‘by design’, then the program is wrong but you cannot say it has a bug.”
There will be more about Steve and cartoons later.
On his Oracle Security Weblog, Pete Finnigan asks, Is Oracle Security getting better or in other words “is Oracle Security good enough?”, a question posed on the Oracle Security forum not long ago. Pete surveys the responses. The consensus seems to be that it’s better and still improving, but still falling short.
On Oracle Rant, Thomas Roach has a thorough howto on restoring a database from Veritas NetBackup, a follow-up to his popular piece from May on RMAN and Veritas NetBackup integration.
Jonathan Lewis and his readers throw some light a finer point of bind capture, on his Oracle Scratchpad. “For a long time I had assumed that [v$sql_bind_capture] was capturing the bind values used when the optimizer peeked at the variables when it wanted to produce an execution plan. But Iâ€™ve just seen a note on the OTN forums . . . that highlighted my error.” It’s quite subtle, as you’ll see.
Pythian’s Riyaj Shamsudeen delved into tuning latch contention: cache-buffers-chain latches.
On the AskDba.org Weblog, Saurabh Sood offers a solid introduction to an 11g New Feature: Database Replay, beginning, “In critical production systems, before making changes to the environment . . . lots of testing is required . . . to know the effect of those changes and this usually takes lots of time to test for the stability and effects of the change. In Oracle 11g this thing is simplified and a new feature ‘database replay’ is added which can be used to simulate the production workload on a test system.”
Oracle Open World 2008 is on the horizon, so bloggers are talking. Eddie Awad has taken over the Blogger Meetup from Mark Rittman, and posts the coordinates for this year’s get-together.
Dan Norris announces the IOUG RAC Attack event, taking place on August 4-5, in Chicago. It is co-sponsored by the Oracle RAC SIG and the IOUG.
On this very blog, MySQL maven Sheeri Cabral announces that she and Dan will be presenting at OOW on the subject of how to contribute to the community. That’s going to be very worthwhile, I predict.
OSCON concluded last Friday, and there were a lot of MySQL people there. Barton’s Blog has some photo highlights from OSCON, including stills of the MySQL vs. Postgres wrestling match (literally) with contenders Monty “The Dolphin” Widenius and Josh “The Elephant” Berkus. No spoilers. Giuseppe Maxia posts a link to his pics in the comments.
On Open Sources, Zack Urlocker has video of Tim O’Reilly’s interview with Monty and Brian Aker, which dealt, in part, with Drizzle.
The View from 25B has a little fairy tale: “Once upon a time a small village of developers decided to create a database. . . . One day everything changed. A giant from a faraway technology land of perpetual sun approached the villagers . . . One very brave villager decided to take the database back to what they had originally intended . . . He stripped away all of the fancy features they had just added to be able to compete with the giants. As soon as he completed his task, clouds began to form and a light drizzle started to fall.” Again, no spoilers.
On the MySQL Performance Blog, Baron Schwartz opines on the #1 mistake hosting providers make for MySQL servers. “Most hosting providers â€” even the big names â€” continue to install 32-bit GNU/Linux operating systems on 64-bit hardware. This is a serious mistake.” Lots of comment offered too.
Arnold Daniels lays out an alternative way of EAV modeling. “EAV describes a method of saving a large set of attributes, only some of which apply to an individual entity. Normally you would create a table, with a row for each entity and save each attribute in a column. With EAV you save each attribute as a row.” Arnold and his readers discuss the pros and cons of this methodology.
A couple posts on tools. Kevin Burton looks at debugging MySQL applications with wireshark. international geographic introduces Ghetto Profile. Giuseppe debuts his new large test database and test suite.
A couple DB2 items now. Vincent McBurney has been Tooling Around in the IBM InfoSphere, and gives us his Big IBM Information on Demand 2008 Conference Preview.
Susan Visser says that it’s party time, the occasion being DB2′s 25th anniversary. The item links to the birthday party pages for IDUG 2008 Europe, and Information On Demand Global Conference.
Finally, here’s Steve Karam and his new cartoon, The Adventures of Ace, DBA. Hey Steve, if you make a success of this, maybe you’ll become an xdba!
Gotta run! If you think I’ve missed any good blogs, please add them in the comments. See you in a week!